Agreed about preference to approval. My problem with IRV isn't about strategic voting though; I think predicting the outcome is too complex to be concerned about that. The worry that I have is that a voting method that encourages candidates to move public opinion away from them will lead to weird outcomes and undermines public trust. I think the 2000 election had an example of this in plurality voting, where republicans campaigned for Nader hoping to split the vote. In IRV I worry that we'll see similar effects, where candidates campaign for opponents because they think it will give them a tactical advantage. Aesthetically that doesn't sound "fair" to me, and my worry is that weird behavior like that undermines trust in the system.